Automatic Data Clustering by Genetic Algorithm Validated by Fuzzy Intercluster Hostility Index

Abstract

One must have a prior knowledge about the optimal number of clusters in a data set before clustering. Without having information regarding the exact nature of the underlying data distribution, the determination of optimal number of clusters in an unlabeled data set is not an easy task. Genetic algorithms (GAs) is known as a randomized search and optimization technique guided by the principles of evolution and natural genetics and efficient enough to handle this type of problems. An application of GA to the automatic clustering of the large unlabeled multidimensional data sets is narrated in this article. A fuzzy intercluster hostility index is proposed in this GA based clustering algorithm and employed to determine the optimal number of clusters from unlabeled multidimensional data sets. Comparative studies with the Automatic Clustering Differential Evolution (ACDE) algorithm shows superior result when these two algorithms are applied on two well-known real-life multidimensional data sets.